Population Pharmacokinetics and Exposure–Response Analyses of Vepdegestrant, a First‐in‐Class PROteolysis‐TArgeting Chimera Estrogen Receptor Degrader

Abstract

Population pharmacokinetic (PK) and exposure–response analyses were performed to characterize the PK and exposure–response relationships of vepdegestrant, a first‐in‐class, oral PROteolysis‐TArgeting Chimera estrogen receptor degrader. Population PK and exposure–response analyses for safety utilized data from the first‐in‐human study (ARV‐471‐mBC‐101, NCT04072952) and the registrational VERITAC‐2 study (NCT05654623), which included patients with ER‐positive, human epidermal growth factor receptor 2 (HER2)‐negative advanced breast cancer. Safety endpoints of clinical interest were evaluated using logistic regression and included grade ≥3 treatment‐emergent adverse events (TEAEs) and TEAEs of any grade (arthralgia, fatigue, nausea, aspartate aminotransferase or alanine aminotransferase elevations, anemia, and neutrophil count decreased). Exposure–efficacy analysis included patients with estrogen receptor‐1 (ESR1)‐mutated, ER‐positive, HER2‐negative advanced breast cancer from the VERITAC‐2 vepdegestrant arm only. Progression‐free survival (PFS) as assessed by blinded independent central review, the primary efficacy endpoint in VERITAC‐2, was assessed via Cox proportional hazards modeling. Vepdegestrant PK was described by a two‐compartment model with linear elimination and sequential zero‐, first‐order absorption. None of the evaluated covariates significantly influenced the disposition of vepdegestrant. There was no statistically significant relationship between vepdegestrant exposure and any of the evaluated safety endpoints across 30‐500 mg total daily doses. In patients with ESR1‐mutated, ER‐positive, HER2‐negative advanced breast cancer treated with vepdegestrant 200 mg once daily in the VERITAC‐2 study, exposure was not a statistically significant predictor of PFS. Overall, integrated analyses adequately characterized the PK of vepdegestrant, with no clinically meaningful covariate effects. No exposure–response relationships were identified between vepdegestrant exposure and efficacy or safety outcomes.

Introduction

Vepdegestrant (Veppanu) is a first‐in‐class, oral PROteolysis‐TArgeting Chimera estrogen receptor (ER) degrader that directly harnesses the ubiquitin‐proteasome system, the primary intracellular protein‐disposal mechanism. ¹ Unlike selective ER degraders, vepdegestrant simultaneously binds ERs and an E3 ligase, forming a ternary complex that results in direct polyubiquitination of ERs and its degradation by the proteasome. ² , ³ , ⁴ Vepdegestrant is approved by the US Food and Drug Administration for the treatment of adults with ER‐positive, human epidermal growth factor receptor 2 (HER2)‐negative, estrogen receptor‐1 (ESR1)‐mutated advanced or metastatic breast cancer at the recommended dose of 200 mg orally once daily with food. ⁵ , ⁶

In a recent randomized Phase 3 study (funded by Pfizer and Arvinas Estrogen Receptor; VERITAC‐2, NCT05654623), vepdegestrant oral daily doses of 200 mg with food significantly prolonged progression‐free survival (PFS) compared with fulvestrant in the subgroup with ESR1 mutations (5.0 months [95% CI, 3.7 to 7.4 vs 2.1 months [95% CI, 1.9 to 3.5]) but not in the full patient population (3.8 months [95% CI, 3.7 to 5.3] vs 3.6 months [95% CI, 2.6 to 4.0]). ⁵ Vepdegestrant showed a favorable safety profile with mainly low‐grade adverse events (AEs). ⁵ Treatment‐emergent adverse events (TEAE) of grade 3 or higher occurred in 23.4% and 17.6% of the patients in the vepdegestrant group and the fulvestrant group, respectively. The most common AEs of any grade included, but were not limited to, fatigue, alanine aminotransferase (ALT) level increased, aspartate aminotransferase (AST) level increased, nausea, anemia, and neutrophil count decreased. ⁵ AEs that led to treatment discontinuation occurred in 2.9% and 0.7% of the patients in the vepdegestrant group and the fulvestrant group, respectively. ⁵

Vepdegestrant is administered orally with food to enhance absorption during clinical development. Maximum plasma concentration (C_max) occurs approximately 6 h after a single oral 200 mg dose. ⁶ , ⁷ Apparent volume of distribution is 764 L (coefficient of variation 26%). ⁶ , ⁷ Vepdegestrant is primarily metabolized through direct sulfation via multiple sulfotransferase isoforms and oxidation via cytochrome P450 (CYP) 3A4 and to a lesser extent by CYP2C8, CYP2C9, and CYP3A5. ⁶ The plasma terminal half‐life is approximately 56 h. ⁷ Following a single 200 mg oral dose of [¹⁴C] radiolabeled vepdegestrant, approximately 68% of the dose was recovered in feces (18% unchanged), and 1.5% was recovered in urine (<0.02% unchanged). ⁶ Vepdegestrant exposures (area under plasma concentration–time curve and C_max) increased in an approximately dose proportional manner over the dose range of 100 to 500 mg daily. ⁶

Here, we present the first population pharmacokinetic (PK) and exposure–response analyses for safety and efficacy that supported the approval of vepdegestrant. The current analyses evaluate the impact of covariates, such as demographic factors, organ function, disease characteristics, body weight, and laboratory values, on the disposition of vepdegestrant and evaluate the relationships between vepdegestrant exposure and efficacy and safety outcomes in patients with advanced breast cancer.

Methods

Study Populations and Clinical Data

All trials were conducted in accordance with principles derived from international guidelines, including the Declaration of Helsinki, the Council for International Organizations of Medical Sciences International Ethical Guidelines, applicable International Council for Harmonisation guidelines for Good Clinical Practice, and other applicable laws and regulations. The trial protocols were approved by an institutional review board and ethics committee at each site. All patients provided written informed consent.

The scope of our modeling analyses is limited to studies in which vepdegestrant was evaluated as monotherapy. The analyses included data from two clinical studies, the first‐in‐human, Phase 1/2 study ARV‐471‐mBC‐101 (NCT04072952, referred to as first‐in‐human Study 101 thereafter) and/or registrational Phase 3 study VERITAC‐2 (NCT05654623), in patients with ER‐positive, HER2‐negative advanced breast cancer. An overview of these two studies is presented in Table S1.

First‐in‐human Study 101 was a multicenter, open‐label Phase 1/2 study that included 3 + 3 dose escalation (Phase 1) and dose expansion (Phase 2) in the monotherapy setting. The study also included a combination cohort with palbociclib, which is beyond the scope of the present analyses and therefore is not described in this manuscript. Eligible patients had ER‐positive, HER2‐negative metastatic, recurrent, or locally advanced unresectable breast cancer previously treated with one cyclin‐dependent kinase 4/6 inhibitor and ≥2 (Phase 1) or ≥1 (Phase 2) endocrine regimens (NCT04072952). ⁸ In Phase 1, the primary objectives were to determine the maximum tolerated dose and recommended Phase 2 doses. Efficacy and safety of the recommended Phase 2 doses (200 and 500 mg oral daily doses, respectively) were assessed in Phase 2. PK samples for vepdegestrant were collected pre‐ and post‐dose as described in Table S1. Plasma concentrations were measured using validated, sensitive, high‐performance liquid chromatography‐tandem mass spectrometric (LC‐MS/MS) methods (refer to Supplemental Information for details).

VERITAC‐2 was a pivotal Phase 3, open‐label, randomized study. The study design and main efficacy and safety results have been previously published. ⁵ Eligible patients were randomly assigned in a 1:1 ratio to receive vepdegestrant at a dose of 200 mg orally once daily with food (Arm A) or fulvestrant at a dose of 500 mg administered intramuscularly on Day 1 and Day 15 of cycle 1 and on Day 1 of subsequent cycles (Arm B). The primary endpoint was PFS as assessed by blinded independent central review (BICR) among patients with ESR1 mutations and all patients who were randomly assigned. PK samples for vepdegestrant were collected pre‐ and post‐dose as described in Table S1. Plasma concentrations were measured using validated, sensitive, high‐performance LC‐MS/MS methods (refer to Supplemental Information for details).

Population PK Analysis

Population PK analysis was conducted using data from pooled studies including first‐in‐human Study 101 (Phase 1/2, total daily doses of 30 to 500 mg) and VERITAC‐2 (Arm A, 200 mg daily dose). A nonlinear mixed effect modeling approach was used with first‐order conditional estimation with interaction estimation in NONMEM 7.5.0 (ICON Development Solutions, Ellicott City, MD). Several absorption and elimination models were evaluated to characterize the PK of vepdegestrant in addition to models evaluating the inclusion of allometric scaling.

Prespecified covariates were evaluated in the population PK analysis based on mechanistic plausibility, exploratory analysis, and scientific interest. These covariates are presented in Table S2 and include demographic characteristics, laboratory measurements, organ function categories, and disease characteristics. Missing continuous covariates were imputed as the median value across the analysis population, and missing categorical covariates were imputed as the mode if the missing covariate was present in <10% of the analysis population. Perl‐speaks‐NONMEM version 5.3.0 (Uppsala University, Uppsala, Sweden) was used to perform the stepwise covariate model (SCM), where model selection was based on a log‐likelihood ratio test. Prespecified cutoff P‐values of .05 and .001 were used for the forward selection and backward elimination steps, respectively. In the case of highly correlated covariates, the covariate that was more clinically relevant was selected.

Goodness‐of‐fit of different models was evaluated through model diagnostics including the following: individual and population‐predicted concentrations compared to observed concentrations; conditional weighted residuals compared to time and observed concentrations along with the distribution of conditional weighted residuals; as well as prediction‐corrected visual predictive check. Bootstrapping was performed for the base and final model using 1000 replicates, and both ɳ‐shrinkage and ε‐shrinkage were evaluated to assess the validity of using post hoc individual parameter estimates for model diagnostics. ⁹ , ¹⁰ , ¹¹

Empirical Bayes estimates and the recorded dosing history for each patient were utilized to simulate exposures to be used in the subsequent exposure–response analyses.

Exposure–Response for Safety

The relationships between vepdegestrant plasma exposure and selected safety endpoints of clinical interest were investigated using binomial logistic regression (R version 4.2.1) for the pooled first‐in‐human Study 101 and VERITAC‐2 study. Safety endpoints included TEAE grade ≥3, specific TEAEs of any grade (arthralgia, diarrhea, fatigue [cluster term], nausea, and vomiting), and specific TEAEs of any grade based on laboratory test (AST or ALT level increased, anemia, and neutrophil count decreased). Linear and log‐transformed model‐predicted vepdegestrant exposures were used including maximum concentration in the first cycle (C_{max, cycle1}) and average vepdegestrant concentration in the first cycle (C_{avg, cycle1}). For each endpoint (event or nonevent), the most severe event was considered as the event. For AE endpoints derived from laboratory values, the event was defined as the most severe post‐baseline grade that worsened from baseline.

In the base model, exposures were tested as predictors for the occurrence of an event in separate univariate logistic regression analyses. The exposure with the largest change in deviance was selected for inclusion in the base model regardless of statistical significance. Covariates in scope are described in Table S2 (with footnote specifications) and were tested for inclusion in the full and final models with a linear parameterization. If any of the covariates were highly correlated (r≥ 0.6), then only 1 of the correlated covariates was tested further based on changes in deviance in separate univariate analyses.

Final model development started with the full model, containing the parameters from the base model and any additional covariates under consideration. The full model was then subjected to a stepwise backward elimination procedure, where covariates and exposure were subjected to elimination. The elimination process stopped when the removal of any of the remaining parameters resulted in a change of deviance equivalent to P<.01, and the model was considered final. The Hosmer–Lemeshow test was used for the goodness‐of‐fit. ¹² , ¹³ The receiver‐operating characteristic curve and concordance statistic were used to assess the model predictive performance.

Exposure–Response for Efficacy

The relationship between vepdegestrant plasma exposure (C_{avg, cycle1}) and efficacy endpoint, PFS as assessed by BICR, was evaluated in patients with ESR1‐mutated (per interactive response technology) advanced breast cancer who received vepdegestrant monotherapy in the registrational VERITAC‐2 Study (referred to as ESR1 mutation subgroup thereafter in this manuscript). This analysis included those treated with vepdegestrant from the VERITAC‐2 study and had population PK‐derived exposures. PFS data were first plotted using Kaplan–Meier survival curves and stratified by exposure quartiles to examine any potential relationships. PFS data were then analyzed using Cox proportional hazards model in R (version 4.2.1), with terms for exposure measures and other prognostic factors (covariates). This model estimates the hazard, or risk of an event occurring at a given time, as a function of a baseline hazard and the influence of covariates.

Univariate Cox proportional hazard analyses were conducted to screen for potential covariates, and a multivariate Cox proportional hazard model was used to estimate the effect of vepdegestrant exposure in the presence of other covariates. Significant covariates identified from the univariate analysis were then simultaneously included in a multivariate Cox proportional hazard model. Vepdegestrant exposure was retained in each step of multivariate analyses. Next, insignificant covariates (Wald test P‐value > .05) were removed from the model, and this process was repeated until all the covariates in the multivariate model all had Wald test P‐value ≤.05.

The proportional hazards assumption for the Cox proportional hazard models of efficacy endpoints was checked graphically using a plot of scaled Schoenfeld residuals along a smooth curve. The x‐axis was the transformed time, while the y‐axis gave an estimate of the time‐dependent coefficient beta(t). Systematic departures from a horizontal line were indicative of nonproportional hazards.

Results

Population PK Analysis

A descriptive summary of baseline participant characteristics from the pooling of first‐in‐human Study 101 and VERITAC‐2 is presented in Table S2. A total of 452 patients from first‐in‐human Study 101 (N = 147) and VERITAC‐2 (N = 305) who received vepdegestrant monotherapy and had available vepdegestrant concentrations were included in the pooled analyses. Analysis included 3625 vepdegestrant PK samples, with 4.1% of samples excluded due to being below the limit of quantification.

A two‐compartment base model with linear elimination and sequential zero‐, first‐order absorption was found to be the most appropriate model describing the population PK of vepdegestrant, where zero‐order input of drug into the gut compartment was characterized by its duration (D1), followed by first‐order absorption into the central compartment. The final model parameter estimates with bootstrap results are presented in Table 1.

Table 1.

Final Vepdegestrant Population PK Model Parameter Estimates

Parameter	Value	RSE (%)	Shrinkage (%)	Bootstrap Median	Bootstrap 95% CI
CL/F (L/h)	11.8	3.8	‐	11.8	(10.9‐12.6)
Vc/F (L)	478	5.3	‐	469	(349‐528)
Q/F (L/h)	2.56	17.5	‐	2.68	(1.81‐5.56)
Vp/F (L)	3062	45.6	‐	2940	(975‐7369)
Ka (per h)	0.80	20.8	‐	0.72	(0.35‐1.22)
Residual error	0.47	5.0	‐	0.47	(0.42‐0.52)
D1 (h)	2.78	9.7	‐	2.67	(1.98‐3.38)

IIV	Value	CV (%)	Shrinkage (%)	Bootstrap Median	Bootstrap 95% CI
IIV on CL/F	0.19	44	12.0	0.20	(0.15‐0.26)
IIV on CL/F and Vc/F correlation	0.09	30	‐	0.10	(0.03‐0.23)
IIV on Vc/F	0.25	50	39.0	0.26	(0.16‐0.50)
OFV	−771	‐	‐	−869	(−1503 to −219)

CI, confidence interval; CL/F, apparent clearance; CV, coefficient of variation; D1, duration (estimated parameter for the zero‐order absorption duration); IIV, inter‐individual variability; h, hours; Ka, first‐order absorption rate constant; L, liter; OFV, objective function value; PK, pharmacokinetic; RSE, residual standard error; Q/F, apparent inter‐compartmental clearance; V_c/F, apparent volume of distribution; V_p/F, apparent peripheral volume of distribution.

In the SCM analysis, relevant covariates, such as baseline weight, categorical hepatic impairment, categorical renal impairment, baseline creatinine clearance, Eastern Cooperative Oncology Group performance status, albumin, race, age, sex, and ESR1 mutation, were tested on apparent clearance (CL/F); covariates, such as age, baseline weight, and sex, were tested on apparent central volume (V_c/F) along with exploratory covariates evaluated on both as presented in Table S2. No covariate was included in the forward addition of the SCM based on prespecified SCM criteria, and therefore the base model was the final model.

The final model adequately captured the totality of vepdegestrant concentration data (Figure 1 and Figure S1). There is a slight underprediction of the absorption profile on Day 1. The shrinkage on V_c/F is high (38.9%, Table 1); however, goodness‐of‐fit plots for conditional weighted residuals and population prediction indicated a lack of impact from the high shrinkage on V_c/F, with data randomly spread around the zero‐reference line and line of unity, respectively. The prediction‐corrected visual predictive checks described the central tendency of the data especially where the majority of concentration data were observed (Figure 1). Additional plots of simulated vepdegestrant concentrations by renal and hepatic function are provided in Figure S2.

Figure 1.

Prediction‐corrected visual predictive checks for the final vepdegestrant population PK model for (A) First‐in‐human Study 101 overall, (B) First‐in‐human Study 101 Day 1, (C) First‐in‐human Study 101 Day 15, and (D) VERITAC‐2 Study overall. The observed data are represented by blue circles and the lines (median, 5th, 95th percentiles). The blue‐shaded ribbon represents the 90% prediction interval for the median population prediction. The red‐shaded ribbons represent the 90% prediction interval for the 5th and 95th population predictions. PK, pharmacokinetic.

Exposure–Response for Safety

A total of 452 patients in the safety dataset with post hoc PK values were included in the exposure–response analysis for safety. The incidence of selected safety endpoints is presented in Table 2. The incidence of TEAE diarrhea and vomiting of any grade were each <10% and therefore not modeled. Vepdegestrant exposure stratified by event status (yes or no) are presented in Figure S3, displaying substantial overlap in exposures from patients with and without the AE of interest. In addition, quantile plots illustrating the relationship between vepdegestrant exposure quartiles and AE endpoints are presented in Figures S4 and S5. Overall, AE incidences were generally similar across exposure quartiles, with no clear trend observed between incidence of AEs and vepdegestrant exposure quartiles.

Table 2.

Observed Incidence of the Safety Endpoints Evaluated in the Pooled Exposure–Response Analysis for Safety

Safety Endpoints	VERITAC‐2 (N = 305)	First‐in‐Human Study 101 (N = 147)	Total (N = 452)
TEAE grade ≥3	68 (22.3)	42 (28.6)	110 (24.3)
AST or ALT increased any grade	102 (33.4)	69 (46.9)	171 (37.8)
Anemia any grade	67 (22)	45 (30.6)	112 (24.8)
Arthralgia any grade	33 (10.8)	35 (23.8)	68 (15)
Diarrhea any grade	18 (5.9)	19 (12.9)	37 (8.2)
Fatigue any grade a	81 (26.6)	67 (45.6)	148 (32.7)
Nausea any grade	42 (13.8)	47 (32)	89 (19.7)
Neutrophil count decreased any grade	69 (22.6)	43 (29.3)	112 (24.8)
Vomiting any grade	20 (6.6)	21 (14.3)	41 (9.1)

ALT, alanine aminotransferase; AST, aspartate aminotransferase; TEAE, treatment‐emergent adverse event.

^a Fatigue (cluster term) was evaluated.

There were no significant exposure–response relationships identified between vepdegestrant exposure and any of the seven safety endpoints evaluated in the final analysis. Final model results are shown in Table S3. Exposures were not retained in the final models as significant covariates for any of the models. Some safety analyses had significant covariate relationships in the final multivariate model; however, covariate effects were not subsequently interpreted in the absence of an exposure–response relationship (Table S3).

Exposure–Response for Efficacy

A total of 132 patients in the ESR1 mutation subgroup from the vepdegestrant monotherapy arm of the VERITAC‐2 study were included in the exposure–response analysis for PFS (as assessed by BICR). A descriptive summary of baseline characteristics for these patients is shown in Table S4. Kaplan–Meier plots of PFS stratified by C_{avg, cycle1} exposure quartiles (Figure 2) demonstrated that C_{avg, cycle1} quartiles do not show a clear relationship with PFS in the analysis population.

Figure 2.

Kaplan–Meier plot of PFS for patients with ESR1 mutation receiving vepdegestrant monotherapy in VERITAC‐2 Study stratified by exposure quartiles (C_{avg, cycle1}). PFS as assessed by BICR was evaluated. BICR, blinded independent central review; C_{avg, cycle1}, average concentration in cycle 1; ESR1, estrogen receptor‐1; PFS, progression‐free survival.

The final multivariate Cox proportional hazard model (Table 3) showed that C_{avg, cycle1} does not have a statistically significant association with PFS. Individual and global Schoenfeld test P‐values and visual examination of the residuals (Figure S6) showed that the model was satisfactory, with no deviation from the assumption of constant hazard ratio over time. Measurable disease at baseline had a statistically significant association with shorter PFS (Table 3). However, this covariate effect was not interpreted in the absence of an exposure–response relationship.

Table 3.

Final Multivariate Cox Proportional Hazard Analysis for PFS in the ESR1 Mutation Subgroup From the VERITAC‐2 Study Vepdegestrant Monotherapy Arm

Cox Proportional Hazard Analysis	Covariates	Coefficient	Hazard Ratio	95% CI for Hazard Ratio	Pr(>|z|)
	Baseline measurable disease	0.973	2.65	1.46‐4.81	0.0014
	Log vepdegestrant Cavg, cycle1	−0.056	0.95	0.44‐2.04	0.887

BICR, blinded independent central review; C_{avg, cycle1}, average concentration in cycle 1; CI, confidence interval; ESR1, estrogen receptor‐1; PFS, progression‐free survival; Pr(>|z|), Wald test P‐value.

Note: PFS as assessed by BICR was used as the efficacy endpoint.

Discussion

This publication presents the first comprehensive population PK, exposure–safety, and exposure–efficacy analyses for vepdegestrant, a first‐in‐class PROteolysis‐TArgeting Chimera ER degrader. Our population PK analysis is based on data from 452 patients who received vepdegestrant monotherapy across total oral daily doses ranging from 30 to 500 mg in the pooled first‐in‐human Study 101 and the VERITAC‐2 registrational study. Vepdegestrant PK data were well described by a two‐compartment model with sequential zero‐, first‐order absorption and linear elimination. All key model structural parameters were estimated with reasonable precision, and the final model described the plasma vepdegestrant concentration data well. Overall, these evaluations and model performance support the use of model‐predicted exposures for the subsequent exposure–response analyses.

Based on the population PK analysis, none of the evaluated covariates (Table S2), such as age, sex, body weight, race, ESR1 mutation status, and ethnicity, had a clinically significant impact on the PK of vepdegestrant. In particular, body weight was not shown to be a significant covariate in the model, which was further supported by the finding that models with allometric scaling did not perform better than models without allometric scaling. The lack of race impact on the disposition of vepdegestrant is also in line with findings from the Japan Phase 1 study ¹⁴ and the China Phase 1 study. ¹⁵ In addition, vepdegestrant exposures were similar when comparing participants with mild and moderate renal impairment to those with normal renal function (Figure S2). Only one participant had severe renal impairment (Table S2). These findings aligned with the minimal contribution of renal clearance to vepdegestrant elimination. ¹⁶ Likewise, vepdegestrant exposures were comparable between patients with normal hepatic function and those with mild hepatic impairment, with only one participant with moderate hepatic impairment (Figure S2 and Table S2).

Different pooling strategies were employed for the exposure–response analyses for safety and efficacy to meet the distinct objectives of each evaluation. For exposure–safety analyses, data from both the first‐in‐human Study 101 and the registrational VERITAC‐2 study were integrated. By pooling across a broad range of evaluated doses (30 to 500 mg total daily doses) in the vepdegestrant monotherapy setting, we utilized a wide dose range to identify potential exposure–safety relationships. Conversely, the exposure–response analysis for efficacy was restricted to patients with ESR1 mutation who received the recommended vepdegestrant 200 mg daily dose within the well‐controlled, randomized VERITAC‐2 registrational study, in which PFS as assessed by BICR served as the primary study endpoint.

Exposure–response analyses for seven safety endpoints of interest did not identify any statistically significant exposure–safety relationships across the 30 to 500 mg dose range, corresponding to approximately 0.15 to 2.5 times the recommended 200 mg dose evaluated in the VERITAC‐2 study. Final exposure–response analyses for TEAE grade ≥3, arthralgia, fatigue, nausea, neutrophil count decreased, and AST or ALT increased only included significant covariates, and exposure was eliminated in the final analysis (Table S3). Notably, the primary objective of these exposure–safety analyses was to examine the relationship between exposure and safety endpoints while adjusting for potential confounders. Therefore, significant covariates identified in these analyses without exposure retained in the final model have limited value and should be interpreted cautiously due to the lack of a clear relationship between exposure and safety endpoints. Overall, these observations from the exposure–response analyses were consistent with the favorable safety profile and tolerability of vepdegestrant demonstrated throughout clinical development. Furthermore, these analyses also suggest that changes in vepdegestrant exposures are unlikely to impact the incidence of the AEs evaluated here.

In the exposure–response analysis for efficacy, our evaluation of the model‐predicted vepdegestrant exposure C_{avg, cycle1} and PFS in the ESR1 mutation subgroup from the VERITAC‐2 study did not identify vepdegestrant exposure as a significant predictor of PFS outcome. Kaplan–Meier analyses stratified by exposure quartiles (Figure 2) showed no apparent relationship between C_{avg, cycle1} and PFS, a finding corroborated by the final multivariate Cox proportional hazard analysis, which revealed no statistically significant associations (Table 3). Examination of baseline characteristics by exposure quartiles suggested no major imbalances of prognostic factors or major trends of imbalances between exposure quartiles (trend from low exposure quartile [Q1] to high exposure quartile [Q4] or vice versa) that would impact the analysis outcome. In addition, we evaluated a broad range of covariates including demographic characteristics and baseline disease‐related factors, with exposure to further assess the potential exposure–efficacy relationship in the presence of covariates. Measurable disease at baseline was the only significant covariate identified in the final model, with its presence at baseline being associated with shorter PFS. However, the lack of a clear and significant relationship between exposure and PFS limited the interpretative value of this covariate and therefore was not further elucidated in the absence of a significant exposure–response relationship. Overall, the exposure–response analysis for efficacy demonstrated that patients benefited similarly across the exposure range associated with the recommended vepdegestrant 200 mg oral daily doses in the vepdegestrant monotherapy test arm from the registrational VERITAC‐2 study.

This is the first population PK and exposure–response analyses of vepdegestrant integrating data across Phase 1/2 to Phase 3 studies that contained multiple dose levels. The PK of vepdegestrant was adequately described by the final population PK model, and the population PK results indicate that vepdegestrant disposition is not meaningfully affected by the covariates evaluated, supporting a favorable PK profile. No clinically meaningful relationship was observed between vepdegestrant exposure and PFS at the recommended 200 mg once‐daily dose. In addition, vepdegestrant exposure was not associated with safety endpoints across the evaluated dose range of 30 to 500 mg total daily doses. Collectively, these analyses provided a comprehensive characterization of the PK and assessment of the exposure–response relationships for vepdegestrant.

Author Contributions

Derek Z. Yang and Brian Jermain performed the primary data analyses. All authors contributed to the analysis planning, interpretation of findings, and manuscript preparation and review. All authors have given approval for the final version to be published.

Conflicts of Interest

Derek Z. Yang, Joanna C. Masters, Hechuan Wang, Lana Tran, Weiwei Tan, Kimberly C. Lee, and Brian Jermain are employees of Pfizer Inc. Yuanyuan Zhang is an employee of Arvinas Operations, Inc.

Funding

The clinical studies on which the analyses were based were sponsored by Arvinas Estrogen Receptor, Inc., in collaboration with Pfizer Inc, or by Pfizer Inc, in collaboration with Arvinas Estrogen Receptor, Inc.

Supporting information

Supporting Information

Data Availability Statement

Requests for data should be directed to the corresponding author. Scientifically sound proposals will be reviewed and approved at the discretion of the study sponsors (Pfizer Inc, and Arvinas Estrogen Receptor, Inc.). Any patient‐level data will be anonymized, and study documents will be redacted to protect the privacy of trial participants.